Abstract
Although much research has demonstrated a relationship between negative life events and depressive symptoms, relatively little research has examined the mechanisms that may mediate this relationship. The theories of Blatt (1974), Bowlby (1980), and Gilbert (1992) each propose proximal predictors of depression. In accordance with these theories, this study examined the relationships among perceived losses in self-worth and interpersonal relationships, anaclitic (dependent) and introjective (self-critical) mood states, and depressive symptoms following a significant negative life event. A sample of 172 undergraduate students completed measures of depressive symptoms and depressive vulnerability factors and retrospectively described the worst period of their lives. They also rated the extent to which the events surrounding this worst period affected their self-worth and their relationships with close others. Structural equation modeling demonstrated that the effect of a perceived loss of self-worth on depressive symptoms was fully mediated by both introjective and anaclitic mood states, whereas the effect of a perceived loss of interpersonal relationships on depressive symptoms was fully mediated by an anaclitic mood state. Additionally, perceived losses of self-worth showed a stronger effect on introjective mood in highly self-critical individuals. Findings highlight the importance of perceived losses in both self-worth and interpersonal domains in response to adverse life events and suggest pathways through which perceived losses may affect depressive symptoms.
Acknowledgements
We thank Allison Kelly and Michelle Leybman for their helpful comments and suggestions on previous drafts of this article. We also thank Dr. Heungsun Hwang for his statistical insights. Finally, we thank our reviewers for their insightful comments on previous versions of this article.
Notes
1. Analyses were recomputed using a BDI score that did not include three items (Items 3, 5, and 6, tapping failure, self-worth, and self-hatred, respectively) that potentially overlapped with introjective mood items. Results did not change substantially, and all significance values for all analyses were unchanged. Because of this, we report results calculated with the full BDI total score.
2. Analyses were recomputed without bootstrapping. Here the direct path from threats to self-worth to depressive symptoms approached significance (β = .101, p = .067).
3. An alternative model was fit to the data in which depressive symptoms predicted mood states, which, in turn, predicted perceived threats to self-worth and relationships. That is, paths in the initial hypothesized model were reversed. After deleting nonsignificant paths, the model yielded an acceptable fit to the data. However, it was not a better fit to the data, according to a chi-square difference test, χ2(2, N = 172) = 1.20, p = ns. Thus, the hypothesized model was preferred.
4. An alternative model was fit to the data in which mood states predicted congruent threats, which then predicted depressive symptoms. That is, our threats variables were tested as mediators of the relationships between mood states and depressive symptoms. This model yielded a relatively poor fit to the data, and threats to self-worth no longer predicted depressive symptoms. This model was also a worse fit to the data compared with our final hypothesized model, according to a chi-square difference test, χ2(2, N = 172) = 11.121, p < .01.